From Shadow AI to Strategic Implementation

As finance teams increasingly explore and advocate for AI tools in their daily work, organizations can no longer postpone developing a structured approach to AI adoption. The question has shifted from "Should we implement AI?" to "How do we implement AI systematically across our organization?" 

In this issue, we're tackling this challenge head-on by providing a comprehensive framework for department and company-wide AI adoption. While individual team members may already be experimenting with AI tools, sustainable implementation requires a coordinated organizational approach that addresses security, compliance, and scalability. 

I'll outline four distinct approaches to organizational AI implementation - from activating AI features in your existing enterprise software to developing comprehensive, custom AI solutions. 

Plus, we'll examine the latest developments in enterprise AI platforms and practical strategies for managing AI artifacts across teams. 

The Balanced View: Exploring AI Implementation Options for Teams and Companies

In this edition, we're diving deep into what it truly means to "implement AI" in the world of finance. The term can mean different things for different organizations, but it essentially revolves around leveraging technology to improve efficiency, reduce errors, and empower teams to make more informed decisions. Let’s explore four distinct ways AI can be adopted by finance departments and professional services firms, ranging from simple integration to transformative solutions.

1. Leveraging AI Features in Existing Software

This is the easiest and fastest way to get started with AI. Many of the software applications you’re already using have built-in AI features designed to improve efficiency and support decision-making.

Why Start Here: This option ranks first because it’s low risk and easy to implement. Vendors are typically responsible for compliance, which means fewer regulatory hurdles for your organization. By activating these existing features, you can achieve quick efficiency gains without significant investments or new integrations.

You might not know it, but your favorite accounting software already has built-in AI features; you only need to turn them on. Many companies offer complimentary training and webinars to their users, but if you want to fully leverage AI potential, I recommend putting your team members through training on AI basics first, such as my AI Blend Workshop.

Below is a list of software with AI features that are already available:

• QuickBooks

• SAP Concur

• Oracle NetSuite

• Xero

• Sage Intacct

• BlackLine

• Workday Adaptive Planning

• Microsoft Dynamics 365 Finance

• Zoho Books

• FreshBooks

...and that's not all of it.

I will explore this further in next week's edition, but for now, here's your homework: Check the software you've been using for ages to find out if it has added AI features, what they are, and how to access them.

2. Subscribing to Enterprise-Level Public AI Models

The next level involves subscribing to enterprise-level public large language models (LLMs) like OpenAI’s GPT or Anthropic’s Claude. These models bring advanced AI capabilities to your organization, including sophisticated language understanding, summarization, and predictive insights. With enterprise subscriptions, you gain features like advanced security settings, better data handling controls, and more powerful functionality.

Why Rank This Second: While these models offer cutting-edge AI features that can be tailored to your needs, they require careful handling of sensitive data. Security and compliance are crucial at this stage, as financial data is often highly sensitive. With proper safeguards in place, however, this step can significantly enhance the capabilities of finance teams, especially for generating insights, drafting reports, and automating client communication.

Finance teams are often overloaded, stretched thin, and asked to do more with less. An enterprise subscription to a public LLM and proper training can be a powerful solution to this challenge.

3. Developing Custom AI Assistants 

Custom AI assistants or bots are tailored solutions that can be designed to automate specific processes within your finance department. For example, a custom bot might handle expense approvals, automate data entry, or perform complex reconciliations. These solutions are built specifically for your organization's unique needs, giving you greater control over how AI is applied.

Why This is Third: Custom AI assistants allow for greater customization and are especially valuable when standard AI tools can’t address unique or complex workflows. However, they involve more complexity compared to using existing software features or subscribing to public models. Developing these solutions requires internal or external expertise, and the time investment is significantly higher. Despite this, custom bots can provide a real competitive edge by automating key processes that are unique to your firm.

In my AI Blend Workshops, we focus on identifying the processes that are ideal candidates for AI automation. We conduct collaborative exercises with company leaders to discover the most significant pain points, followed by an ROI analysis to determine the potential value of automating those processes. From there, we develop a tailored roadmap that outlines the implementation of the most impactful AI assistants, ensuring the solutions are well-aligned with the organization's unique needs.

4. Implementing Comprehensive AI-Based Software Solutions

The most advanced approach involves implementing or developing sophisticated AI-driven systems that become an integral part of your operations. These solutions can automate specific areas, such as accounts payable, accounts receivable, budgeting, or supply chain management, while seamlessly integrating with your existing software.

For organizations seeking a truly transformative solution, a full-scale AI-driven ERP can be implemented, incorporating AI features across all modules—from CRM to HR to Finance—enabling a unified, intelligent approach to managing business processes.

Why Rank This Last: Implementing comprehensive AI-based solutions is the most complex and costly option, requiring significant resources, both in terms of financial investment and time. However, once your team has gained experience through the earlier steps, this level of AI implementation can bring transformational change, fundamentally improving how your department or firm operates.

Security, compliance, and an incremental approach are critical to successful AI adoption in finance. Financial data is sensitive, and the regulatory environment is complex—any implementation of AI must prioritize these aspects. Starting small, by leveraging existing AI features and gradually progressing through more complex solutions, allows your organization to build expertise while managing risks effectively.

AI adoption cannot be achieved sustainably from the bottom up—it requires leadership commitment to invest in the right technologies, establish policies, and train employees. Proper vendor due diligence, compliance procedures, and team education are essential steps to ensure a smooth AI adoption process that safeguards sensitive data while driving efficiency and innovation.

Practical Tip of the Week: Sharing AI Artifacts Across Platforms

AI-generated content can be extremely valuable when shared effectively across your team, particularly for collaborative projects. Let's compare how Claude, ChatGPT, and Perplexity allow you to share artifacts and projects and which is the safest approach.

Claude (Anthropic)

Anthropic's Claude features one of the most sophisticated artifact-sharing systems I've encountered. The platform's dedicated "Artifacts" feature goes beyond simple output storage; it provides a comprehensive system for managing various types of content, such as code snippets, text documents, graphics, diagrams, and website designs. This is particularly valuable for finance teams as it can handle complex code repositories, interactive visualizations, and more. Artifacts are easy to share using the publish button, which generates a link that can be shared with anyone. However, it's worth noting that access control is limited unless you're on a Team plan, and you can only share the end product, not the entire conversation.

The Team plan gives much wider collaboration capabilities, including Shared Project Spaces, Shared Activity Feeds, and advanced security and data access settings.

ChatGPT

ChatGPT takes a different approach. On the free or individual plan, sharing options are relatively basic, often limited to copying and pasting outputs or exporting chat histories. Custom GPTs can be published and made accessible to other users, but these are pre-built chats tailored to specific tasks rather than true collaboration tools.

However, the platform is great when it comes to Team subscription offerings. The "Dedicated Workspaces" feature allows teams to create organized environments for specific projects, such as budget planning or financial analysis. The Teams subscription also includes enhanced data control and security, which are particularly relevant for handling sensitive financial information, ensuring that AI collaboration aligns with financial data security standards.

Perplexity

Perplexity's "Spaces" feature makes it an interesting player in this space, focusing on real-time collaboration. It is a shared research environment where team members can contribute simultaneously. The platform's approach to security is noteworthy—they've built-in controls to keep sensitive financial data private while still leveraging AI capabilities. Users can customize privacy levels and exclude certain files from AI training to protect proprietary data.

I am a big fan of Claude's artifacts, especially for their versatility in managing different types of content. However, if you're dealing with any kind of confidential data, it's crucial to upgrade to a Team plan to ensure proper security and access controls.

ChatGPT, while powerful, has limitations in team collaboration compared to other platforms. On the other hand, Perplexity's new "Spaces" feature is rapidly becoming my go-to tool for collaborative research.

Below is a brief comparison of sharing features for the three platforms (for paid plans).

AI News of the Week: Claude's Latest Upgrades

This week, Claude received a significant upgrade with the introduction of the Claude 3.5 Sonnet and Claude 3.5 Haiku models, both enhancing coding abilities, reasoning skills, and performance, while offering low latency. Additionally, the new Computer Use Capability (Beta) allows Claude to navigate interfaces like a human, promising to automate repetitive tasks and streamline workflows, making Claude a more versatile tool for various applications.

Claude's recent upgrades include enhanced data analytics capabilities, making it an even more powerful tool for financial teams. Key improvements include:

• Handling Large Volumes of Data

• Agentic Tool Use for Data Analysis

• Stronger Reasoning and Problem-Solving

Although I am a fan of Claude, the recent upgrades haven't completely won me over—particularly because it still lacks the ability to process Excel files and has rather low data usage limits. ChatGPT continues to lead in data processing. However, the pace of improvement across models is fascinating, and it's clear that the competition for leadership in this space is still very much alive. Let's keep an eye on how things evolve!

Closing Thoughts

The journey to AI adoption in finance doesn't have to be overwhelming. As we've explored in this issue, success often comes from starting small and scaling thoughtfully. Whether you're just beginning to explore AI features in your existing software or ready to implement comprehensive AI solutions, the key is to move forward with intention and balance. 

Remember, the goal isn't to adopt AI for AI's sake but to enhance your team's capabilities and drive real business value. Start where you are, use what you have, and build your AI capabilities incrementally. The most successful implementations I've seen aren't necessarily the most ambitious - they're the ones that align closely with business needs and prioritize security and compliance from day one. 

We Want Your Feedback!

This newsletter is for you, and we want to make it as valuable as possible. Please reply to this email with your questions, comments, or topics you'd like to see covered in future issues. Your input shapes our content!

Want to dive deeper into balanced AI adoption for your finance team? Or do you want to hire an AI-powered CFO? Book a consultation! 

Did you find this newsletter helpful? Forward it to a colleague who might benefit!

Until next Tuesday, keep balancing!

Anna Tiomina 
CFO & AI Expert